It looks like you are using an ad blocker. That's okay. Who doesn't? But without advertising revenue, we can't keep making this site awesome. Click the link below for instructions on disabling adblock.
Welcome to the Newschoolers forums! You may read the forums as a guest, however you must be a registered member to post.
Register to become a member today!
Quote from the article -- "At this week’s winter X Games in Aspen, Colo., judges will carefully cast their eyes over the athletes, as they have since the first edition of the games in 1997, scrutinizing each spin, flip and grab. But this year, another party will be weighing in as well: an A.I. judge. During the snowboard SuperPipe competitions, video cameras will capture the moves of each athlete, and artificial intelligence will take that information, do what artificial intelligence does and provide a score for the routine."
lmao
I feel like if you can't differentiate both ways spins, grabs, and rail tricks, you probably shouldn't judge the games lool
just hire 4 local pros with decent media training and have at it
I doubt that the "AI" they're using is actually AI and I also doubt they're actually using it at all, every company is lying about AI since it's a pretty solid way to garner more investment. "Our new AI tool will help us to...." and it's just an "if" statement in Excel or some shit lol
Farmville420I doubt that the "AI" they're using is actually AI and I also doubt they're actually using it at all, every company is lying about AI since it's a pretty solid way to garner more investment. "Our new AI tool will help us to...." and it's just an "if" statement in Excel or some shit lol
You nailed it really, "AI" has been around for decades, its just algorithms, we just have more powerful computers now that make it a bit more usable.
They will prob be using some sort of video based AI that they have trained with all of the xgames footage from the past (plus whatever else) and I bet it can identify tricks pretty well but "judging" those tricks is just click bait.
eheathYou nailed it really, "AI" has been around for decades, its just algorithms, we just have more powerful computers now that make it a bit more usable.
They will prob be using some sort of video based AI that they have trained with all of the xgames footage from the past (plus whatever else) and I bet it can identify tricks pretty well but "judging" those tricks is just click bait.
Yeah I think it will be used exactly like that. Record the run, identify the tricks, and compare the tricks to other runs that included them and the final score of said run. It might generate scores, but my assumption is that it's being used to make judging more consistent, if a run historically scored an 87.5 but the judges all agreed on 95.5, it can help them decide "well this run was very similar to an 87.5 two years ago, maybe we should re-evaluate our scores.
That's just my speculation because like you said, 0% chance a computer is actually just giving scores autonomously. Even with real AI there is still a human element there to make sure there's nothing funky going on.
this is a huge nothingburger. most ai can't even generate an image of a full wineglass since they're only trained on images of partially full glasses. how are they supposed to judge new tricks?
Farmville420Yeah I think it will be used exactly like that. Record the run, identify the tricks, and compare the tricks to other runs that included them and the final score of said run. It might generate scores, but my assumption is that it's being used to make judging more consistent, if a run historically scored an 87.5 but the judges all agreed on 95.5, it can help them decide "well this run was very similar to an 87.5 two years ago, maybe we should re-evaluate our scores.
That's just my speculation because like you said, 0% chance a computer is actually just giving scores autonomously. Even with real AI there is still a human element there to make sure there's nothing funky going on.
Not to mention that most judging, at least in skiing, the scores vary based on the runs. If people are throwing down hard, landing tricks NBDs and putting together crazy runs, the AI would just give them all 99s or something.
Judging in these events isn't absolute, the scores don't correlate to tricks or combos, it's all based on the other runs. They give scores based on how they perform compared other runs, so this sort of AI judging will likely never work.
They're likely using this or something very similar. Considering they have labelled data from every X-Games and other competitions, it's really not a bad idea.
Style tricks would be edge cases that could easily added in as well. Theres no reason that a model wouldnt be able to pick up on highly scored tricks with lower rotations.
IMO this could be a step forward to better scoring overall. I imagine skate 3 type rating system.
Remember that very few of these complex AI systems will replace humans, at least for a while. Humans, especially in "art" areas, will always exist.
**This post was edited on Jan 24th 2025 at 5:21:19pm
Lmao, as someone whose company was contracted by X Games to do this exact same thing while they controlled the graphic output last year, I can tell you that this is all a farce to raise money for their new "X Games League" especially after they ran a PR campaign for this.
If they had a "real time AI or AI Judge", they would have used it where it actually matters, when the trick drops instead of after the run some pre-planned replay thats done by an owl with a score that has an extra few digits to make it look AI. Also, all methods that have been proposed or thought of in this thread absolutely do not work or stand a chance when your analyzing video at night with riders wearing black.
Basically, this is a part of the Grift Games for the owners
eheathyou better go back in time, we've been using "AI" for decades, it just wasn't called AI.
totally right but in some cases it's being pushed on us before it's better than what it's replacing and that's pretty frustrating. for example some AI google results are unbelievably bad
SofaKingSicktotally right but in some cases it's being pushed on us before it's better than what it's replacing and that's pretty frustrating. for example some AI google results are unbelievably bad
Google results is your worst example of it not being better, seriously...?
What about the drones who are programmed to launch and kill targets with no human guidance? What about cars that "drive themselves" but will choose the driver in the cars safety over people on the sidewalk. What about nuclear arms that are controlled by AI?
-skian-What about the drones who are programmed to launch and kill targets with no human guidance? What about cars that "drive themselves" but will choose the driver in the cars safety over people on the sidewalk. What about nuclear arms that are controlled by AI?
-skian-Google results is your worst example of it not being better, seriously...?
What about the drones who are programmed to launch and kill targets with no human guidance? What about cars that "drive themselves" but will choose the driver in the cars safety over people on the sidewalk. What about nuclear arms that are controlled by AI?
i mean, i just brought up the first example that came to mind. everyone uses google and it's gotten noticeably worse. the AI result up top often reads like you asked an extremely confident but stupid 10 year old to google something for you
SofaKingSicktotally right but in some cases it's being pushed on us before it's better than what it's replacing and that's pretty frustrating. for example some AI google results are unbelievably bad
lol not a doomer, I am just pointing out that AI is needing a bit of a check. The folks developing AI are crying out saying the innovation needs to slow down and be considered more.
SofaKingSicki mean, i just brought up the first example that came to mind. everyone uses google and it's gotten noticeably worse. the AI result up top often reads like you asked an extremely confident but stupid 10 year old to google something for you
Also yes I do totally agree haha and sometimes that one song shows up in a spotify radio that reallly does not belong. Not saying you are wrong at all, just saying there are so many scary examples.
